N2: PrOJect 3 Ovarian cancer is a complex disease and it is likely that multiple genes and environmental factors contribute, alone and in combination, to its etiology. While the independent effects of common genetic polymorphisms and of the accepted reproductive/lifestyle risk factors are an important part of the etiolgic puzzle, the puzzle will not be complete until the role of multi-locus effects (gene-gene interactions) and interactions between reproductive/lifestyle factors and genotype are systematically explored. The goal of this project is to implement a systematic evaluation of GxG and GxE effects that applies the most advanced analytic procedures available to the most comprehensive data set on ovarian cancer susceptibility built to date. In particular, we plan to join the individual-level genetic data at each phase of this study, assembled and maintained by Project 1, with the core epidemiolgic data maintained by the Ovarian Cancer Association Consortium (OCAC) to create an analysis data set with more than 10,000 cases and 10,000 controls informative for genetic and environmental interaction effects. Compilation of the epidemiologic data is underway. The specific aims are as follows: 1) To evaluate gene-gene interactions using both agnostic and pathway-based approaches for each of the four largest histological subtypes;2) to evaluate gene-environment interactions, for 14 epidemiologic factors, using both agnostic and pathwaybased approaches for each of the four largest histological subtypes;3) to create a comprehensive model for absolute risk of developing ovarian cancer that incorporates family history (BRCA1/2 status), other accepted genetic and environmental factors, and those discovered (Projects 1 and 3) and functionally characterized (Project 2) by the current study. Aims 1 and 2 share a common two-level analysis plan: a screen comprised of existing complementary methods followed by model-based evaluation of the notable interaction effects that emerge from the screen. The screen's purpose is to evaluate and coarsely prioritize as many and as diverse a set of multi-locus or multivariable GxE effects for more formal and computationally demanding multinomial response, multivariate adjusted analysis. The Aim 3 risk model will stratify patients according to whether their risk of ovarian cancer is driven by familial susceptibility (BRCA1/2) or common epidemiological and genetic risk factors. Model building for Aim 3 will focus on specifying the latter component and incorporating G, GxG and GxE effects identified in this study.